3666
Langmuir 1996, 12, 3666-3670
CF4 Plasma Treatment of Asymmetric Polysulfone Membranes J. Hopkins and J. P. S. Badyal* Department of Chemistry, Science Laboratories, Durham University, Durham DH1 3LE, England Received January 12, 1996. In Final Form: April 29, 1996X
CF4 glow discharge treatment of asymmetric polysulfone membranes has been investigated by X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), and gas permeability measurements. Oxygen and nitrogen gas permeability and permselectivity through the polysulfone substrate are found to be strongly influenced by the plasma-processing parameters.
Introduction Permeability plays an important role in many technological areas including packaging,1 gas separation membranes,2,3 protein filtration,4 and controlled drug release devices.5 Polymeric gas separation membranes with high permeability and enhanced permselectivity (separation factor) are widely sought after. Normally there exists a trade-off between these two performance criteria. Good permselectivity requires a substrate which is rigid, crystalline, and cross-linked, whereas any amorphous regions encourage penetrant mobility through the polymer matrix. Membranes can be categorized as homogeneous (or dense), asymmetric, and composite types. A homogeneous membrane consists of one material with the same type of structure throughout. In contrast, an asymmetric membrane possesses a dense skin layer supported by a porous matrix of the same material; the main advantage of using an asymmetric membrane is that it allows high throughput, due to the underlying “spongelike” matrix acting as a mechanical support offering little resistance toward the flow of gas, whilst the skin layer behaves as the selective part of the membrane. Composite membranes are based on a combination of materials with vastly different chemical structures.6 In principle, the gas separation performance of a membrane can be improved by the chemical modification of the membrane structure itself.7 A variety of methods can be used for the modification of membranes; these include UV grafting of functionalities,8 fluorination,9 photo-oxidation,10 ion beam irradiation,11 and plasma * To whom correspondence should be addressed. X Abstract published in Advance ACS Abstracts, July 1, 1996. (1) Comyn, J., Ed. Polymer Permeability; Elsevier: Amsterdam, 1985. (2) Lee, S. H.; Kim, J.-J.; Kim, U. Y. J. Appl. Polym. Sci. 1993, 49, 539. (3) Zhou, W.; Gao, X.; Lu, F. J. Appl. Polym. Sci. 1994, 51, 855. (4) Kesting, R. E. Synthetic Polymeric Membranes. A Structural Perspective, 2nd ed.; Wiley: New York, 1985; Chapter 10. (5) Osada, Y.; Nakagawa, T. Membrane Science and Technology Marcel Dekker: New York, 1992; Chapter 12. (6) Staude, E.; Breitbach, L. J. Appl. Polym. Sci. 1991, 43, 559. (7) Dal, M. M.; Tan, C.; Uiver, M. D.; Tweddle, T. A. J. Appl. Polym. Sci. 1994, 54, 783. (8) Nystrom, N.; Jarveien, P. J. Membr. Sci. 1991, 60, 275. (9) Mohr, J. M.; Paul, D. R.; Pinnau, I.; Koros, W. K. J. Membr. Sci. 1991, 56, 77. (10) Meier, I. K.; Langsam, M.; Cklotz, H. J. Membr. Sci. 1994, 94, 195. (11) Lin, X.; Dolveck, Y.; Boiteaux, G.; Escoubes, M.; Monchanin, M.; Dupin, J. P.; Davenas, J. J. Appl. Polym. Sci. 1995, 55, 99.
S0743-7463(96)00038-8 CCC: $12.00
treatment.12,13 Non-isothermal plasma techniques can be employed to deposit a thin layer of plasma polymer onto the membrane to form a composite structure,14-17 or alternatively chemical functionalites can be grafted directly onto the membrane surface.18 The latter offers a number of advantages including the fact that it is quicker and cheaper, produces less waste, and does not significantly lower the gas flux through the membrane. In this study, the CF4 plasma modification of asymmetric polysulfone membrane surfaces has been investigated as a function of electrical discharge power and treatment time using X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), and gas permeability measurements. Experimental Section Asymmetric polysulfone membranes were prepared by casting polysulfone (Udel P3500, Union Carbide) films from a 20% by weight dimethylacetamide (99%+, Lancaster) solution onto a glass substrate at ambient temperature, followed by immersion in deionized water. High-purity carbon tetrafluoride (99.7%, Air Products) gas was used for the electrical discharge treatments. The dense skin side of a small piece of asymmetric polysulfone membrane was plasma treated in a cylindrical glass reactor (4.5 cm diameter, 515 cm3 volume, base pressure of 1.5 × 10-3 mbar, with a leak rate better than 2.0 × 10-3 cm3 min-1) enclosed in a Faraday cage. This was fitted with a gas inlet, a Pirani pressure gauge, and a 27 L min-1 two-stage rotary pump attached to a liquid nitrogen cold trap. A matching network was used to inductively couple a copper coil (4 mm diameter, 13 turns, spanning 9-18 cm from the gas inlet) wound around the reactor to a 13.56 MHz radio frequency (rf) source. All joints were greasefree. Gas flow and leak rates were calculated by assuming ideal gas behavior.19 A typical experimental run comprised initially scrubbing the reactor with detergent, rinsing with isopropyl alcohol, and oven drying; this was followed by a 60 min highpower (50 W) air plasma cleaning treatment. Next, the reactor (12) Stancell, A. F.; Spencer, A. T. J. Appl. Polym. Sci. 1972, 16, 1505. (13) Masouoka, T.; Iwatsubo, T.; Mizoguchi. J. Appl. Polym. Sci. 1992, 46, 311. (14) Matsuyama, H.; Kanya, A.; Teramaoto, M. J. Appl. Polym. Sci. 1994, 51, 689. (15) Lin, X.; Chen, J.; Xu, J. J. Membr. Sci. 1994, 90, 81. (16) Inagaki, N.; Tasaka, S.; Murata, T. J. Appl. Polym. Sci. 1989, 38, 1869. (17) Clarotti, G.; Schue, F.; Sledtz, J.; Geckeler, K. E.; Gopel, W.; Orsetti, A. J. Membr. Sci. 1991, 61, 289. (18) Wang, Y. J.; Chen, C. H.; Yeh, M.; Suie, G. H.; Yu, B. C. J. Membr. Sci. 1990, 53, 275. (19) Ehrlich, C. D.; Basford, J. A. J. Vac. Sci. Technol. 1992, A10, 1.
© 1996 American Chemical Society
CF4 Plasma Treatment of Polysulfone Membranes
Langmuir, Vol. 12, No. 15, 1996 3667
Figure 1. C(1s) XPS spectra of (a) clean polysulfone and (b) CF4 plasma treatment of polysulfone (10 W, 15 min). was opened up to the atmosphere, a strip of membrane was inserted into the center of the rf coils, and then the system was evacuated back down to its original base pressure of 1.5 × 10-3 mbar. Subsequently CF4 gas was introduced into the reaction chamber at a pressure of 2 × 10-1 mbar and a flow rate of approximately 1.9 cm3 min-1 (i.e. at least 99.6% of the reactor contents). The reaction zone was purged with CF4 for 5 min, and then the glow discharge was ignited. Upon completion of plasma treatment, the rf generator was switched off, and CF4 gas was allowed to flow through the reactor for a further 5 min prior to venting to atmospheric pressure. Each sample was characterized immediately after electrical discharge treatment by X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), and gas permeability measurements. A Kratos ES300 electron spectrometer equipped with a Mg KR X-ray source (1253.6 eV) and a concentric hemispherical analyzer was used for XPS surface analysis. Photoemitted core level electrons were collected at a takeoff angle of 30° from the substrate normal, with electron detection in the fixed retarding ratio (FRR, 22:1) mode. XPS spectra were accumulated on an interfaced PC computer and curve fitted using a Marquardt minimization algorithm. Instrument performance was calibrated with respect to the gold 4f7/2 peak at 83.8 eV with a full-widthat-half-maximum (fwhm) of 1.2 eV. Photoelectron ionization cross-section sensitivity factors for unit stoichiometry were taken as being C(1s):O(1s):S(2p):F(1s) equals 1.00:0.55:0.54:0.53. XPS was used to check the cleanliness of the polysulfone substrate and for the absence of any surface-active inorganic additives. Gross and experimental errors were calculated for each surface modification. Mass spectrometric sampling devices20 have previously been used to evaluate the permeability of common elastomers.21,22 In this study, the dense side of the asymmetric polysulfone membrane was treated with a CF4 plasma and placed into a permeability probe.23 This consisted of two drilled-out stainless steel flanges with a copper gasket to ensure a leak-tight seal, and an electron microscope grid to provide mechanical support to the membrane. The probe was coupled to the sample preparation chamber of a Kratos ES300 X-ray photoelectron spectrometer (base pressure of 2 × 10-10 mbar) by means of a stainless steel tube inserted through a rotatable gate valve. The treated side of each sample film was exposed to a gas pressure of 800 mbar. High-purity nitrogen (BOC, 99.995%) and oxygen (BOC, 99.6%) gases were used. The permeant pressure passing through the asymmetric membrane was monitored by an ultrahigh vacuum ion gauge (Vacuum Generators, VIG 24). A quadrupole mass spectrometer (Vacuum Generators SX200) (20) Westover, L. B.; Tou, J. C.; Mark, J. H. Anal. Chem. 1974, 46, 568. (21) Laurenson, L.; Dennis, N. T. M. J. Vac. Sci. Technol. 1985, A3, 1707. (22) Tou, J. C.; Rulf, D. C.; DeLassus, P. T. Anal. Chem. 1990, 62, 592. (23) Barker, C. P.; Kochem, K.-H.; Revell, K. M.; Kelly, R. S. A.; Badyal, J. P. S. Thin Solid Films 1995, 259, 46.
Figure 2. Gas permeation following CF4 plasma treatment of the dense side of an asymmetric polysulfone membrane as a function of power (at a fixed time of 15 min): (a, top) gas permeability; (b, bottom) O2/N2 selectivity. interfaced to a PC computer was used to follow compositional analysis of the permeant species. The quadrupole mass spectrometer’s response per unit pressure was calculated by introducing each gas in turn into the chamber and recording the corresponding mass spectrum at a predetermined pressure of 5 × 10-6 mbar (taking into account ion gauge sensitivity factors24 ). The mean equilibrium permeant partial pressure (MEPPP) of each gas was measured in the steady-state flow regime.25 Atomic force microscopy offers structural characterization of surfaces in the 10-4 to 10-10 m range without the prerequisite of special sample preparation (e.g. metalization). A Digital Instruments Nanoscope III atomic force microscope was used to examine the topographical nature of the polysulfone surface prior to and after electrical discharge exposure. All of the AFM images were acquired in air using the Tapping mode26 and are presented as unfiltered data. This technique employs a stiff silicon cantilever oscillating at a large amplitude near its resonance frequency (several hundred kilohertz). The root mean square amplitude is detected by an optical beam system. A large root mean square amplitude is used to overcome the capillary attraction of the surface layer, whilst the high oscillation frequency allows the cantilever to strike the surface many times before being displaced laterally by one tip diameter. These features offer the advantage of low contact forces and no shear forces. The probe scanning direction was carefully chosen in order to avoid tip-induced artifacts.
Results (a) X-ray Photoelectron Spectroscopy. The elemental C:S:O ratio for untreated polysulfone membrane obtained using XPS was found to be 82.4 ( 0.1%:4.1 ( 0.4%:13.6 ( 0.2%; this is in good agreement with the (24) Pressure measurement technical information, Vacuum Generators Limited. (25) Crank, J., Park, G. S., Eds. Diffusion in Polymers; Academic Press: London, 1968; Chapter 1. (26) Zhong, Q.; Inniss, D.; Kjoller, K.; Elings, V. B. Surf. Sci. 1993, 290, L688.
3668
Langmuir, Vol. 12, No. 15, 1996
Hopkins and Badyal
Figure 3. Gas permeation following CF4 plasma treatment of the dense side of an asymmetric polysulfone membrane as a function of time (at a fixed power of 10 W): (a, top) gas permeability; (b, bottom) O2/N2 selectivity.
theoretically predicted values of 84.4%:3.1%:12.5% for the parent structure of bisphenol-A-polysulfone:
The slightly higher concentration of sulfur can be attributed to the preferential orientation of the sulfone group at the polymer surface.27,28 Mg KR1,2 C(1s) XPS spectra were fitted with Gaussian peaks of equal full-width-athalf-maximum (fwhm),29 using a Marquardt minimization computer program (Figure 1a). Carbon atoms attached to hydrogen/carbon, sulfone, and ether groups exhibit C(1s) core level binding energies of 285.0, 285.6, and 286.6 eV, respectively.28-30 Low-energy π-π* shake-up transitions accompanying core level ionization around 291.7 eV were fitted with a Gaussian peak of different fwhm. The O(1s) envelope is a 1:1 doublet with peaks centred at 532.1 and 533.6 eV corresponding to oxygen bonded to sulfur (sulfone groups) and oxygen bonded to carbon in the backbone (ether linkages).27,30 The S(2p1/2,3/2) peak was found to be an unresolved 2:1 doublet at 168.0 eV; this can be taken as being characteristic of a sulfone group rather than a sulfide (163.6 eV) or sulfate (169.3 eV) environment.28,30-32 (27) Clark, D. T.; Dilks, A.; Peeling, J.; Thomas, H. R. Trans. Faraday Soc., Faraday Discuss. 1975, 60, 183. (28) Clark, D. T.; Thomas, H. R. J. Polym. Sci. Polym. Chem. Ed. 1978, 16, 791. (29) Evans, J.; Gibson, J.; Moulder, J.; Hammond, J.; Goretzki, H. Fresenius’ Z. Anal. Chem. 1984, 319, 841. (30) Beamson, G.; Briggs, D. High Resolution XPS of Organic Polymers: The Scienta ESCA300 Database. (31) Yasuda, H.; Marsh, H. C.; Brandt, E. S.; Reilley, C. N. Am. Chem. Soc. Polym. Pr. 1975, 16, 142.
Figure 4. AFM micrographs of asymmetric polysulfone membrane: (a) dense side; (b) porous side.
CF4 plasma treatment of the asymmetric polysulfone membrane resulted in a substantial amount of fluorine incorporation at the surface33 (Figure 1b). This was accompanied by a dramatic change in the C(1s) XPS spectrum with >CF2 functionalities becoming the predominant fluorine moiety: CF at 287.8 eV, CFnCFn at 289.3 eV, CF2 at 291.2 eV, and CF3 at 293.6 eV30 (Figure 1b). The 1:1 O(1s) doublet could no longer be resolved, and the S(2p1/2,3/2) feature was found to be strongly attenuated. During CF4 glow discharge treatment, the elemental composition of the surface was found to quickly approach the limiting values of approximately 48% C, 48% F, 0.8% S, and 4% O at input powers greater than 5 W (for a fixed treatment time of 15 min) and exposure times longer than 5 min (for a fixed input power of 10 W). Therefore within the context of the present study, all of the CF4 plasma-modified membrane substrates contain the same level of chemical modification at the surface. (b) Permeability Measurements. Figure 2 shows the changes in permeant pressure with increasing glow discharge power over a fixed CF4 plasma treatment period of 15 min. There appears to be an initial drop in permeant pressure at low input powers with respect to the permeation characteristics of the untreated asymmetric polysulfone membrane. The permeant pressure then rises at higher powers. A similar trend is observed with longer treatment times at a constant plasma power of 10 W (Figure 3). In both cases, a corresponding improvement in O2/N2 permselectivity was noted at low powers/short (32) Clark, D. T.; Feast, W. J.; Richie, I.; Musgrave, W. K. R.; Modena, M.; Ragazzini, M. J. Polym. Sci., Polym. Chem. Ed. 1974, 12, 1049. (33) Tou, J. C.; Rulf, D. C.; DeLassus, P. T. Anal. Chem. 1990, 62, 592.
CF4 Plasma Treatment of Polysulfone Membranes
Figure 5. AFM micrographs following CF4 plasma treatment of the dense side of an asymmetric polysulfone membrane as a function of power (at a fixed time of 15 min): (a) 5 W; (b) 30 W.
treatment times (150% and 400% enhancement, respectively) which is followed by a drop in relative gas separation performance with rising power/time approaching values measured for the untreated asymmetric polysulfone membrane. (c) Atomic Force Microscopy. The untreated dense skin side of the asymmetric polysulfone membrane exhibits a compact structure whilst 1.27 ( 0.16 µm pore apertures are clearly discernible on the opposite face (Figure 4). A gradual roughening of the surface was observed during the CF4 plasma treatment of the dense skin side of the asymmetric substrate with increasing input powers and exposure times (Figures 5 and 6, respectively) to eventually unveil small holes at the surface. Discussion CF4 molecules can undergo bond cleavage and ionization in the presence of an alternating rf electromagnetic field. The reactive component of a CF4 plasma is reported to be fluorine atoms with a small concentration of CF, CF2, and CF3 radicals (all having vibrational and rotational temperatures approximately equal to room temperature).34-38 This is supported by electron impact experiments with CF4, which indicate that fluorine atoms are the primary dissociative species.39,40 The vacuum ultraviolet compo(34) Truesdale, E. A.; Smolinsky, G. J. Appl. Phys. 1979, 50, 6594. (35) d’Agostino, R.; Cramossa, F.; DeBenedictus, S. Plasma Chem. Plasma Process. 1982, 2, 213. (36) Hopkins, J.; Badyal, J. P. S. J. Phys. Chem. 1995, 99, 4261. (37) Plumb, I. C.; Ryan, K. R. Plasma Chem. Plasma Process. 1986, 6, 205. (38) Edelson, D.; Flamm, D. L. J. Appl. Phys. 1984, 56, 1522. (39) Kay, E. Proc. Int. Ion Eng. Congr.sISIAT 83 & IPAT 83, Kyoto. 1983, 1657.
Langmuir, Vol. 12, No. 15, 1996 3669
Figure 6. AFM micrographs following CF4 plasma treatment of the dense side of an asymmetric polysulfone membrane as a function of time (at a fixed power of 10 W): (a) 5 min; (b) 30 min.
nent of the glow discharge can also lead to electronic excitation of fluorine atoms.41 This abundance of fluorine radicals can graft onto an organic surface via H-substitution and the opening of unsaturated bonds to yield mCF, >CF2, and -CF3 functionalities.36 Fluorination of polymeric membrane surfaces using F2/ noble gas mixtures is also reported to improve the gas separation characteristics of polymeric substrates.9,42-44 It is interesting to note that straightforward fluorination of asymmetric polysulfone membranes by exposure to F2 gas does not show any change in the surface topography upon treatment.9 Therefore the CF4 glow discharge method used in the present study must be a much more reactive medium compared to F2 gas exposure with the added benefit of generating the fluorinating moieties insitu, thereby making the process safer, cleaner, and cheaper. This leads to the formation of a composite membrane structure as depicted in Figure 7. The dense skin layer and the fluorinated layer should exhibit different permeation properties, leading to an overall net change in the gas transport properties of the system. Permselectivity is based upon the ability of polymer matrices to operate as size and shape selective networks (40) Iriyama, Y.; Yasuda, H. J. Polym. Sci., Polym. Chem. Ed. 1992, 30, 1731. (41) Inagaki, N.; Tasaka, S.; Mori, K. J. Appl. Polym. Sci. 1991, 43, 581. (42) Le Roux, J. D.; Paul, D. R.; Kampa, J.; Lagow, R. J. J. Membr. Sci. 1994, 90, 21. (43) Le Roux, J. D.; Paul, D. R.; Kampa, J.; Lagow, R. J. J. Membr. Sci. 1994, 90, 121. (44) Le Roux, J. D.; Teplyakov, V.; Paul, D. R. J. Membr. Sci. 1994, 90, 55. (45) Mohr, J. M.; Paul, D. R.; Taru, Y.; Mlsna, T. E.; Lagow, R. J. J. Membr. Sci. 1991, 55, 149. (46) Aroantskii, A. E.; Vakar, A. K.; Goluber, A. V.; Kraasheninner, E. G.; Liventsov, V. V.; Macherel, S. O.; Rusanor, V. D.; Fridman, A. A. High Energy Chem. 1990, 24, 223.
3670
Langmuir, Vol. 12, No. 15, 1996
Figure 7. Schematic diagram of CF4 plasma treatment of an asymmetric polysulfone membrane.
via chain mobility and packing.47 Fluorination of a polysulfone membrane by CF4 plasma treatment will give rise to not only cross-linking but also rotational hindrance (since the F atom is larger than the hydrogen atom which it replaces), thereby producing a more rigid structure. The observed drop in permeant pressure and improvement in O2/N2 permselectivity at low powers and short treatment times is consistent with the greater affinity of fluorinated functionalities toward molecular oxygen.48,49 The depth of fluorination is controlled by a balance between fluorination and ablation. The F atoms can etch/ablate the polymer42,50 whilst a combination of F atoms and CFx radicals can contribute toward the formation of a fluorinated layer.51 Therefore, beyond a critical threshold power level/treatment time, the CF4 plasma will begin to etch/ (47) Hellums, M. W.; Koros, W. J.; Husk, G. R.; Paul, D. R. J. Membr. Sci. 1989, 46, 93. (48) Koros, W. J.; Fleming, G. K. J. Membr. Sci. 1993, 83, 1. (49) Toshima, N., Ed. Polymers for Gas Separation; VCH Publishers: New York, 1992. (50) Langotagne, B.; Kuttel, O. M.; Wertheimer, M. R. Can. J. Phys. 1991, 69, 202.
Hopkins and Badyal
degrade the dense skin layer of the asymmetric membrane, since it is typically only in the 500-1000 Å range.9,42,43 This can clearly be seen in the AFM micrographs, where higher input powers and longer treatment times give rise to holes appearing within the dense skin to unveil the underlying “spongelike” matrix of the asymmetric membrane, which in turn increases the gas permeability but causes a loss of any enhancement of permselectivity due to the breakdown of the fluorinated skin. There may also be cross-linking in the subsurface region by the vacuum ultraviolet component of the CF4 plasma; this may contribute together with restricted rotational mobility (due to the larger fluorine atoms) toward the initial drop in permeant pressure at low input powers and short treatment times, since a more rigid structure will present a more tortuous pathway for the permeant gases.45,46 Conclusions CF4 plasma treatment of an asymmetric polysulfone membrane gives rise to extensive surface fluorination, which produces an improvement in its inherent O2/N2 permselectivity characteristics. Longer treatment times and higher input powers etch away the dense skin layer, leading to an enhancement in gas throughput combined with a concomitant drop in O2/N2 permselectivity. Acknowledgment. J.H. thanks EPSRC and British Gas for a CASE Studentship and provision of equipment. LA960038I (51) Strobel, M.; Corn, S.; Lyons, C. S.; Korba, G. A. J. Polym. Sci., Polym. Chem. Ed. 1987, 25, 1295.
